Prompt FeedbackOverview
Prompt FeedbackOverview

Prompt Feedback

August 23, 2026

Project Overview

After my first project, I crave feedback on how I can work with AI better. My obvious answer is to improve my prompts. My first attempt relied on Claude giving me feedback on my prompts (Notebook 1). This was flawed because the feedback wasn’t necessarily accurate or based in true best practices. I restarted this work, researching prompting best practices and creating a document of tactics (Notebook 2).

Introduction

I was dissatisfied. I had just told myself I was officially done with my R6Hub project. While I had learned a lot, not having a working product sucked. What I really needed at that point was feedback - how could I make my prompts better?

That gave me an idea - why not use Claude itself to evaluate our chats? It already had the context, and it could probably tell me what worked well and what didn’t.

That initial spark kicked off this larger project. After all, why stay limited to feedback only at a point in time? A longer-term feedback mechanism would help me continually learn and improve. I could create tangible, useful tools that would help me be a better prompter to AI in general.

The notebooks

The detailed, step-by-step write-ups for this project.

  1. Hindsight is 20-20

    Debriefing my flawed approach to compile tactics and a scorecard for prompts. What I did, what I learned, why it was wrong.

  2. Tactics. For real this time

    Researching and compiling prompting tactics from knowledgeable sources. I share a completed tactics document.